Mercurial > emacs
annotate src/character.h @ 109750:0851022edd4b
* xfns.c (Fx_hide_tip): Check FRAME_LIVE_P (f) before calling xg_hide_tooltip.
author | Jan D <jan.h.d@swipnet.se> |
---|---|
date | Wed, 11 Aug 2010 20:47:34 +0200 |
parents | 05e7e7c46ff0 |
children | fda36a325177 |
rev | line source |
---|---|
88363 | 1 /* Header for multibyte character handler. |
2 Copyright (C) 1995, 1997, 1998 Electrotechnical Laboratory, JAPAN. | |
89483 | 3 Licensed to the Free Software Foundation. |
106815 | 4 Copyright (C) 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010 |
88363 | 5 National Institute of Advanced Industrial Science and Technology (AIST) |
6 Registration Number H13PRO009 | |
7 | |
8 This file is part of GNU Emacs. | |
9 | |
94994
29adfc9354e7
Switch to recommended form of GPLv3 permissions notice.
Glenn Morris <rgm@gnu.org>
parents:
91805
diff
changeset
|
10 GNU Emacs is free software: you can redistribute it and/or modify |
88363 | 11 it under the terms of the GNU General Public License as published by |
94994
29adfc9354e7
Switch to recommended form of GPLv3 permissions notice.
Glenn Morris <rgm@gnu.org>
parents:
91805
diff
changeset
|
12 the Free Software Foundation, either version 3 of the License, or |
29adfc9354e7
Switch to recommended form of GPLv3 permissions notice.
Glenn Morris <rgm@gnu.org>
parents:
91805
diff
changeset
|
13 (at your option) any later version. |
88363 | 14 |
15 GNU Emacs is distributed in the hope that it will be useful, | |
16 but WITHOUT ANY WARRANTY; without even the implied warranty of | |
17 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
18 GNU General Public License for more details. | |
19 | |
20 You should have received a copy of the GNU General Public License | |
94994
29adfc9354e7
Switch to recommended form of GPLv3 permissions notice.
Glenn Morris <rgm@gnu.org>
parents:
91805
diff
changeset
|
21 along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>. */ |
88363 | 22 |
23 #ifndef EMACS_CHARACTER_H | |
24 #define EMACS_CHARACTER_H | |
25 | |
88428
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
26 /* character code 1st byte byte sequence |
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
27 -------------- -------- ------------- |
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
28 0-7F 00..7F 0xxxxxxx |
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
29 80-7FF C2..DF 110xxxxx 10xxxxxx |
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
30 800-FFFF E0..EF 1110xxxx 10xxxxxx 10xxxxxx |
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
31 10000-1FFFFF F0..F7 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx |
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
32 200000-3FFF7F F8 11111000 1000xxxx 10xxxxxx 10xxxxxx 10xxxxxx |
89692 | 33 3FFF80-3FFFFF C0..C1 1100000x 10xxxxxx (for eight-bit-char) |
34 400000-... invalid | |
88363 | 35 |
89692 | 36 invalid 1st byte 80..BF 10xxxxxx |
37 F9..FF 11111xxx (xxx != 000) | |
88363 | 38 */ |
39 | |
88428
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
40 /* Maximum character code ((1 << CHARACTERBITS) - 1). */ |
88363 | 41 #define MAX_CHAR 0x3FFFFF |
42 | |
88428
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
43 /* Maximum Unicode character code. */ |
88363 | 44 #define MAX_UNICODE_CHAR 0x10FFFF |
45 | |
88428
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
46 /* Maximum N-byte character codes. */ |
88363 | 47 #define MAX_1_BYTE_CHAR 0x7F |
48 #define MAX_2_BYTE_CHAR 0x7FF | |
49 #define MAX_3_BYTE_CHAR 0xFFFF | |
50 #define MAX_4_BYTE_CHAR 0x1FFFFF | |
51 #define MAX_5_BYTE_CHAR 0x3FFF7F | |
52 | |
90791
0603ad3252c9
(MIN_MULTIBYTE_LEADING_CODE)
Kenichi Handa <handa@m17n.org>
parents:
90533
diff
changeset
|
53 /* Minimum leading code of multibyte characters. */ |
0603ad3252c9
(MIN_MULTIBYTE_LEADING_CODE)
Kenichi Handa <handa@m17n.org>
parents:
90533
diff
changeset
|
54 #define MIN_MULTIBYTE_LEADING_CODE 0xC0 |
0603ad3252c9
(MIN_MULTIBYTE_LEADING_CODE)
Kenichi Handa <handa@m17n.org>
parents:
90533
diff
changeset
|
55 /* Maximum leading code of multibyte characters. */ |
0603ad3252c9
(MIN_MULTIBYTE_LEADING_CODE)
Kenichi Handa <handa@m17n.org>
parents:
90533
diff
changeset
|
56 #define MAX_MULTIBYTE_LEADING_CODE 0xF8 |
0603ad3252c9
(MIN_MULTIBYTE_LEADING_CODE)
Kenichi Handa <handa@m17n.org>
parents:
90533
diff
changeset
|
57 |
88946
233c080b5756
(CHAR_TO_BYTE8): If C is not eight-bit char, call
Kenichi Handa <handa@m17n.org>
parents:
88915
diff
changeset
|
58 /* Nonzero iff C is a character that corresponds to a raw 8-bit |
233c080b5756
(CHAR_TO_BYTE8): If C is not eight-bit char, call
Kenichi Handa <handa@m17n.org>
parents:
88915
diff
changeset
|
59 byte. */ |
233c080b5756
(CHAR_TO_BYTE8): If C is not eight-bit char, call
Kenichi Handa <handa@m17n.org>
parents:
88915
diff
changeset
|
60 #define CHAR_BYTE8_P(c) ((c) > MAX_5_BYTE_CHAR) |
233c080b5756
(CHAR_TO_BYTE8): If C is not eight-bit char, call
Kenichi Handa <handa@m17n.org>
parents:
88915
diff
changeset
|
61 |
88428
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
62 /* Return the character code for raw 8-bit byte BYTE. */ |
88363 | 63 #define BYTE8_TO_CHAR(byte) ((byte) + 0x3FFF00) |
88428
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
64 |
105661
bac26aa40069
Remove leftover table unibyte_to_multibyte_table.
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
103741
diff
changeset
|
65 #define UNIBYTE_TO_CHAR(byte) \ |
bac26aa40069
Remove leftover table unibyte_to_multibyte_table.
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
103741
diff
changeset
|
66 (ASCII_BYTE_P (byte) ? (byte) : BYTE8_TO_CHAR (byte)) |
bac26aa40069
Remove leftover table unibyte_to_multibyte_table.
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
103741
diff
changeset
|
67 |
88428
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
68 /* Return the raw 8-bit byte for character C. */ |
88946
233c080b5756
(CHAR_TO_BYTE8): If C is not eight-bit char, call
Kenichi Handa <handa@m17n.org>
parents:
88915
diff
changeset
|
69 #define CHAR_TO_BYTE8(c) \ |
233c080b5756
(CHAR_TO_BYTE8): If C is not eight-bit char, call
Kenichi Handa <handa@m17n.org>
parents:
88915
diff
changeset
|
70 (CHAR_BYTE8_P (c) \ |
233c080b5756
(CHAR_TO_BYTE8): If C is not eight-bit char, call
Kenichi Handa <handa@m17n.org>
parents:
88915
diff
changeset
|
71 ? (c) - 0x3FFF00 \ |
233c080b5756
(CHAR_TO_BYTE8): If C is not eight-bit char, call
Kenichi Handa <handa@m17n.org>
parents:
88915
diff
changeset
|
72 : multibyte_char_to_unibyte (c, Qnil)) |
88428
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
73 |
95856
f13a77e0e34f
* character.h (CHAR_TO_BYTE_SAFE): New macro.
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
95576
diff
changeset
|
74 /* Return the raw 8-bit byte for character C, |
f13a77e0e34f
* character.h (CHAR_TO_BYTE_SAFE): New macro.
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
95576
diff
changeset
|
75 or -1 if C doesn't correspond to a byte. */ |
f13a77e0e34f
* character.h (CHAR_TO_BYTE_SAFE): New macro.
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
95576
diff
changeset
|
76 #define CHAR_TO_BYTE_SAFE(c) \ |
f13a77e0e34f
* character.h (CHAR_TO_BYTE_SAFE): New macro.
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
95576
diff
changeset
|
77 (CHAR_BYTE8_P (c) \ |
f13a77e0e34f
* character.h (CHAR_TO_BYTE_SAFE): New macro.
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
95576
diff
changeset
|
78 ? (c) - 0x3FFF00 \ |
95858
d719efd8c628
(CHAR_TO_BYTE_SAFE): Cut&paste error.
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
95856
diff
changeset
|
79 : multibyte_char_to_unibyte_safe (c)) |
95856
f13a77e0e34f
* character.h (CHAR_TO_BYTE_SAFE): New macro.
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
95576
diff
changeset
|
80 |
88428
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
81 /* Nonzero iff BYTE is the 1st byte of a multibyte form of a character |
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
82 that corresponds to a raw 8-bit byte. */ |
88363 | 83 #define CHAR_BYTE8_HEAD_P(byte) ((byte) == 0xC0 || (byte) == 0xC1) |
84 | |
89053
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
85 /* If C is not ASCII, make it unibyte. */ |
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
86 #define MAKE_CHAR_UNIBYTE(c) \ |
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
87 do { \ |
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
88 if (! ASCII_CHAR_P (c)) \ |
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
89 c = CHAR_TO_BYTE8 (c); \ |
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
90 } while (0) |
89018
a9f683a73092
(MAKE_CHAR_UNIBYTE, MAKE_CHAR_MULTIBYTE): New macros.
Kenichi Handa <handa@m17n.org>
parents:
88946
diff
changeset
|
91 |
a9f683a73092
(MAKE_CHAR_UNIBYTE, MAKE_CHAR_MULTIBYTE): New macros.
Kenichi Handa <handa@m17n.org>
parents:
88946
diff
changeset
|
92 |
98025
d4e07000ca4c
Fix spelling and wording of comments.
Eli Zaretskii <eliz@gnu.org>
parents:
97817
diff
changeset
|
93 /* If C is not ASCII, make it multibyte. Assumes C < 256. */ |
95576
abcb40dd43a2
(MAKE_CHAR_MULTIBYTE): Check the arg is a (uni)byte.
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
94994
diff
changeset
|
94 #define MAKE_CHAR_MULTIBYTE(c) \ |
105661
bac26aa40069
Remove leftover table unibyte_to_multibyte_table.
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
103741
diff
changeset
|
95 (eassert ((c) >= 0 && (c) < 256), (c) = UNIBYTE_TO_CHAR (c)) |
89018
a9f683a73092
(MAKE_CHAR_UNIBYTE, MAKE_CHAR_MULTIBYTE): New macros.
Kenichi Handa <handa@m17n.org>
parents:
88946
diff
changeset
|
96 |
88428
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
97 /* This is the maximum byte length of multibyte form. */ |
88363 | 98 #define MAX_MULTIBYTE_LENGTH 5 |
99 | |
98025
d4e07000ca4c
Fix spelling and wording of comments.
Eli Zaretskii <eliz@gnu.org>
parents:
97817
diff
changeset
|
100 /* Return a Lisp character whose character code is C. Assumes C is |
89887
028a1f06f612
(LEADING_CODE_LATIN_1_MIN) (LEADING_CODE_LATIN_1_MAX): Delete these
Kenichi Handa <handa@m17n.org>
parents:
89692
diff
changeset
|
101 a valid character code. */ |
88363 | 102 #define make_char(c) make_number (c) |
103 | |
104 /* Nonzero iff C is an ASCII byte. */ | |
105 #define ASCII_BYTE_P(c) ((unsigned) (c) < 0x80) | |
106 | |
107 /* Nonzero iff X is a character. */ | |
108 #define CHARACTERP(x) (NATNUMP (x) && XFASTINT (x) <= MAX_CHAR) | |
109 | |
98025
d4e07000ca4c
Fix spelling and wording of comments.
Eli Zaretskii <eliz@gnu.org>
parents:
97817
diff
changeset
|
110 /* Nonzero iff C is valid as a character code. GENERICP is not used. */ |
88428
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
111 #define CHAR_VALID_P(c, genericp) ((unsigned) (c) <= MAX_CHAR) |
88363 | 112 |
113 /* Check if Lisp object X is a character or not. */ | |
90533 | 114 #define CHECK_CHARACTER(x) \ |
115 CHECK_TYPE (CHARACTERP (x), Qcharacterp, x) | |
88363 | 116 |
89483 | 117 #define CHECK_CHARACTER_CAR(x) \ |
118 do { \ | |
119 Lisp_Object tmp = XCAR (x); \ | |
120 CHECK_CHARACTER (tmp); \ | |
121 XSETCAR ((x), tmp); \ | |
122 } while (0) | |
123 | |
124 #define CHECK_CHARACTER_CDR(x) \ | |
125 do { \ | |
126 Lisp_Object tmp = XCDR (x); \ | |
127 CHECK_CHARACTER (tmp); \ | |
128 XSETCDR ((x), tmp); \ | |
129 } while (0) | |
130 | |
88363 | 131 /* Nonzero iff C is an ASCII character. */ |
132 #define ASCII_CHAR_P(c) ((unsigned) (c) < 0x80) | |
133 | |
134 /* Nonzero iff C is a character of code less than 0x100. */ | |
135 #define SINGLE_BYTE_CHAR_P(c) ((unsigned) (c) < 0x100) | |
136 | |
137 /* Nonzero if character C has a printable glyph. */ | |
138 #define CHAR_PRINTABLE_P(c) \ | |
107056
a6dabb2fc53a
* character.h (CHAR_PRINTABLE_P): Reparenthesize to avoid warning.
Andreas Schwab <schwab@linux-m68k.org>
parents:
106815
diff
changeset
|
139 (((c) >= 32 && (c) < 127) \ |
a6dabb2fc53a
* character.h (CHAR_PRINTABLE_P): Reparenthesize to avoid warning.
Andreas Schwab <schwab@linux-m68k.org>
parents:
106815
diff
changeset
|
140 || ! NILP (CHAR_TABLE_REF (Vprintable_chars, (c)))) |
88363 | 141 |
88428
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
142 /* Return byte length of multibyte form for character C. */ |
88363 | 143 #define CHAR_BYTES(c) \ |
144 ( (c) <= MAX_1_BYTE_CHAR ? 1 \ | |
145 : (c) <= MAX_2_BYTE_CHAR ? 2 \ | |
146 : (c) <= MAX_3_BYTE_CHAR ? 3 \ | |
147 : (c) <= MAX_4_BYTE_CHAR ? 4 \ | |
148 : (c) <= MAX_5_BYTE_CHAR ? 5 \ | |
149 : 2) | |
150 | |
89053
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
151 |
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
152 /* Return the leading code of multibyte form of C. */ |
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
153 #define CHAR_LEADING_CODE(c) \ |
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
154 ((c) <= MAX_1_BYTE_CHAR ? c \ |
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
155 : (c) <= MAX_2_BYTE_CHAR ? (0xC0 | ((c) >> 6)) \ |
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
156 : (c) <= MAX_3_BYTE_CHAR ? (0xE0 | ((c) >> 12)) \ |
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
157 : (c) <= MAX_4_BYTE_CHAR ? (0xF0 | ((c) >> 18)) \ |
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
158 : (c) <= MAX_5_BYTE_CHAR ? 0xF8 \ |
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
159 : (0xC0 | (((c) >> 6) & 0x01))) |
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
160 |
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
161 |
88428
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
162 /* Store multibyte form of the character C in P. The caller should |
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
163 allocate at least MAX_MULTIBYTE_LENGTH bytes area at P in advance. |
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
164 Returns the length of the multibyte form. */ |
88363 | 165 |
166 #define CHAR_STRING(c, p) \ | |
167 ((unsigned) (c) <= MAX_1_BYTE_CHAR \ | |
168 ? ((p)[0] = (c), \ | |
169 1) \ | |
170 : (unsigned) (c) <= MAX_2_BYTE_CHAR \ | |
171 ? ((p)[0] = (0xC0 | ((c) >> 6)), \ | |
172 (p)[1] = (0x80 | ((c) & 0x3F)), \ | |
173 2) \ | |
174 : (unsigned) (c) <= MAX_3_BYTE_CHAR \ | |
175 ? ((p)[0] = (0xE0 | ((c) >> 12)), \ | |
176 (p)[1] = (0x80 | (((c) >> 6) & 0x3F)), \ | |
177 (p)[2] = (0x80 | ((c) & 0x3F)), \ | |
178 3) \ | |
90798
ad70c7654800
(CHAR_STRING): Cast C to unsigned on calling
Kenichi Handa <handa@m17n.org>
parents:
90791
diff
changeset
|
179 : char_string ((unsigned) c, p)) |
88363 | 180 |
89180
1d29c2b108e6
(CHAR_STRING): Call char_string if C is greater than
Kenichi Handa <handa@m17n.org>
parents:
89053
diff
changeset
|
181 /* Store multibyte form of byte B in P. The caller should allocate at |
1d29c2b108e6
(CHAR_STRING): Call char_string if C is greater than
Kenichi Handa <handa@m17n.org>
parents:
89053
diff
changeset
|
182 least MAX_MULTIBYTE_LENGTH bytes area at P in advance. Returns the |
1d29c2b108e6
(CHAR_STRING): Call char_string if C is greater than
Kenichi Handa <handa@m17n.org>
parents:
89053
diff
changeset
|
183 length of the multibyte form. */ |
89038
f60ed671d6e4
(BYTE8_STRING): New macro.
Kenichi Handa <handa@m17n.org>
parents:
89018
diff
changeset
|
184 |
f60ed671d6e4
(BYTE8_STRING): New macro.
Kenichi Handa <handa@m17n.org>
parents:
89018
diff
changeset
|
185 #define BYTE8_STRING(b, p) \ |
f60ed671d6e4
(BYTE8_STRING): New macro.
Kenichi Handa <handa@m17n.org>
parents:
89018
diff
changeset
|
186 ((p)[0] = (0xC0 | (((b) >> 6) & 0x01)), \ |
90051 | 187 (p)[1] = (0x80 | ((b) & 0x3F)), \ |
89038
f60ed671d6e4
(BYTE8_STRING): New macro.
Kenichi Handa <handa@m17n.org>
parents:
89018
diff
changeset
|
188 2) |
f60ed671d6e4
(BYTE8_STRING): New macro.
Kenichi Handa <handa@m17n.org>
parents:
89018
diff
changeset
|
189 |
88363 | 190 |
98025
d4e07000ca4c
Fix spelling and wording of comments.
Eli Zaretskii <eliz@gnu.org>
parents:
97817
diff
changeset
|
191 /* Store multibyte form of the character C in P and advance P to the |
d4e07000ca4c
Fix spelling and wording of comments.
Eli Zaretskii <eliz@gnu.org>
parents:
97817
diff
changeset
|
192 end of the multibyte form. The caller should allocate at least |
d4e07000ca4c
Fix spelling and wording of comments.
Eli Zaretskii <eliz@gnu.org>
parents:
97817
diff
changeset
|
193 MAX_MULTIBYTE_LENGTH bytes area at P in advance. */ |
88363 | 194 |
89180
1d29c2b108e6
(CHAR_STRING): Call char_string if C is greater than
Kenichi Handa <handa@m17n.org>
parents:
89053
diff
changeset
|
195 #define CHAR_STRING_ADVANCE(c, p) \ |
1d29c2b108e6
(CHAR_STRING): Call char_string if C is greater than
Kenichi Handa <handa@m17n.org>
parents:
89053
diff
changeset
|
196 do { \ |
1d29c2b108e6
(CHAR_STRING): Call char_string if C is greater than
Kenichi Handa <handa@m17n.org>
parents:
89053
diff
changeset
|
197 if ((c) <= MAX_1_BYTE_CHAR) \ |
1d29c2b108e6
(CHAR_STRING): Call char_string if C is greater than
Kenichi Handa <handa@m17n.org>
parents:
89053
diff
changeset
|
198 *(p)++ = (c); \ |
1d29c2b108e6
(CHAR_STRING): Call char_string if C is greater than
Kenichi Handa <handa@m17n.org>
parents:
89053
diff
changeset
|
199 else if ((c) <= MAX_2_BYTE_CHAR) \ |
1d29c2b108e6
(CHAR_STRING): Call char_string if C is greater than
Kenichi Handa <handa@m17n.org>
parents:
89053
diff
changeset
|
200 *(p)++ = (0xC0 | ((c) >> 6)), \ |
1d29c2b108e6
(CHAR_STRING): Call char_string if C is greater than
Kenichi Handa <handa@m17n.org>
parents:
89053
diff
changeset
|
201 *(p)++ = (0x80 | ((c) & 0x3F)); \ |
1d29c2b108e6
(CHAR_STRING): Call char_string if C is greater than
Kenichi Handa <handa@m17n.org>
parents:
89053
diff
changeset
|
202 else if ((c) <= MAX_3_BYTE_CHAR) \ |
1d29c2b108e6
(CHAR_STRING): Call char_string if C is greater than
Kenichi Handa <handa@m17n.org>
parents:
89053
diff
changeset
|
203 *(p)++ = (0xE0 | ((c) >> 12)), \ |
1d29c2b108e6
(CHAR_STRING): Call char_string if C is greater than
Kenichi Handa <handa@m17n.org>
parents:
89053
diff
changeset
|
204 *(p)++ = (0x80 | (((c) >> 6) & 0x3F)), \ |
1d29c2b108e6
(CHAR_STRING): Call char_string if C is greater than
Kenichi Handa <handa@m17n.org>
parents:
89053
diff
changeset
|
205 *(p)++ = (0x80 | ((c) & 0x3F)); \ |
1d29c2b108e6
(CHAR_STRING): Call char_string if C is greater than
Kenichi Handa <handa@m17n.org>
parents:
89053
diff
changeset
|
206 else \ |
1d29c2b108e6
(CHAR_STRING): Call char_string if C is greater than
Kenichi Handa <handa@m17n.org>
parents:
89053
diff
changeset
|
207 (p) += char_string ((c), (p)); \ |
88428
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
208 } while (0) |
88363 | 209 |
89180
1d29c2b108e6
(CHAR_STRING): Call char_string if C is greater than
Kenichi Handa <handa@m17n.org>
parents:
89053
diff
changeset
|
210 |
88363 | 211 /* Nonzero iff BYTE starts a non-ASCII character in a multibyte |
212 form. */ | |
213 #define LEADING_CODE_P(byte) (((byte) & 0xC0) == 0xC0) | |
214 | |
88873
7d441bc35e9b
(TRAILING_CODE_P): New macro.
Kenichi Handa <handa@m17n.org>
parents:
88832
diff
changeset
|
215 /* Nonzero iff BYTE is a trailing code of a non-ASCII character in a |
7d441bc35e9b
(TRAILING_CODE_P): New macro.
Kenichi Handa <handa@m17n.org>
parents:
88832
diff
changeset
|
216 multibyte form. */ |
7d441bc35e9b
(TRAILING_CODE_P): New macro.
Kenichi Handa <handa@m17n.org>
parents:
88832
diff
changeset
|
217 #define TRAILING_CODE_P(byte) (((byte) & 0xC0) == 0x80) |
7d441bc35e9b
(TRAILING_CODE_P): New macro.
Kenichi Handa <handa@m17n.org>
parents:
88832
diff
changeset
|
218 |
88428
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
219 /* Nonzero iff BYTE starts a character in a multibyte form. |
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
220 This is equivalent to: |
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
221 (ASCII_BYTE_P (byte) || LEADING_CODE_P (byte)) */ |
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
222 #define CHAR_HEAD_P(byte) (((byte) & 0xC0) != 0x80) |
5eaa8c11ab45
(CHAR_VALID_P): Don't call CHARACTERP.
Kenichi Handa <handa@m17n.org>
parents:
88363
diff
changeset
|
223 |
88363 | 224 /* How many bytes a character that starts with BYTE occupies in a |
225 multibyte form. */ | |
226 #define BYTES_BY_CHAR_HEAD(byte) \ | |
227 (!((byte) & 0x80) ? 1 \ | |
228 : !((byte) & 0x20) ? 2 \ | |
229 : !((byte) & 0x10) ? 3 \ | |
230 : !((byte) & 0x08) ? 4 \ | |
231 : 5) | |
232 | |
233 | |
234 /* The byte length of multibyte form at unibyte string P ending at | |
98025
d4e07000ca4c
Fix spelling and wording of comments.
Eli Zaretskii <eliz@gnu.org>
parents:
97817
diff
changeset
|
235 PEND. If STR doesn't point to a valid multibyte form, return 0. */ |
88363 | 236 |
237 #define MULTIBYTE_LENGTH(p, pend) \ | |
238 (p >= pend ? 0 \ | |
239 : !((p)[0] & 0x80) ? 1 \ | |
240 : ((p + 1 >= pend) || (((p)[1] & 0xC0) != 0x80)) ? 0 \ | |
241 : ((p)[0] & 0xE0) == 0xC0 ? 2 \ | |
242 : ((p + 2 >= pend) || (((p)[2] & 0xC0) != 0x80)) ? 0 \ | |
243 : ((p)[0] & 0xF0) == 0xE0 ? 3 \ | |
244 : ((p + 3 >= pend) || (((p)[3] & 0xC0) != 0x80)) ? 0 \ | |
245 : ((p)[0] & 0xF8) == 0xF0 ? 4 \ | |
246 : ((p + 4 >= pend) || (((p)[4] & 0xC0) != 0x80)) ? 0 \ | |
247 : (p)[0] == 0xF8 && ((p)[1] & 0xF0) == 0x80 ? 5 \ | |
248 : 0) | |
249 | |
250 | |
98025
d4e07000ca4c
Fix spelling and wording of comments.
Eli Zaretskii <eliz@gnu.org>
parents:
97817
diff
changeset
|
251 /* Like MULTIBYTE_LENGTH, but don't check the ending address. */ |
88363 | 252 |
253 #define MULTIBYTE_LENGTH_NO_CHECK(p) \ | |
254 (!((p)[0] & 0x80) ? 1 \ | |
255 : ((p)[1] & 0xC0) != 0x80 ? 0 \ | |
256 : ((p)[0] & 0xE0) == 0xC0 ? 2 \ | |
257 : ((p)[2] & 0xC0) != 0x80 ? 0 \ | |
258 : ((p)[0] & 0xF0) == 0xE0 ? 3 \ | |
259 : ((p)[3] & 0xC0) != 0x80 ? 0 \ | |
260 : ((p)[0] & 0xF8) == 0xF0 ? 4 \ | |
261 : ((p)[4] & 0xC0) != 0x80 ? 0 \ | |
262 : (p)[0] == 0xF8 && ((p)[1] & 0xF0) == 0x80 ? 5 \ | |
263 : 0) | |
264 | |
98025
d4e07000ca4c
Fix spelling and wording of comments.
Eli Zaretskii <eliz@gnu.org>
parents:
97817
diff
changeset
|
265 /* If P is before LIMIT, advance P to the next character boundary. |
d4e07000ca4c
Fix spelling and wording of comments.
Eli Zaretskii <eliz@gnu.org>
parents:
97817
diff
changeset
|
266 Assumes that P is already at a character boundary of the same |
108933 | 267 multibyte form whose end address is LIMIT. */ |
89483 | 268 |
269 #define NEXT_CHAR_BOUNDARY(p, limit) \ | |
270 do { \ | |
271 if ((p) < (limit)) \ | |
272 (p) += BYTES_BY_CHAR_HEAD (*(p)); \ | |
273 } while (0) | |
274 | |
275 | |
276 /* If P is after LIMIT, advance P to the previous character boundary. | |
98025
d4e07000ca4c
Fix spelling and wording of comments.
Eli Zaretskii <eliz@gnu.org>
parents:
97817
diff
changeset
|
277 Assumes that P is already at a character boundary of the same |
108933 | 278 multibyte form whose beginning address is LIMIT. */ |
89483 | 279 |
280 #define PREV_CHAR_BOUNDARY(p, limit) \ | |
281 do { \ | |
282 if ((p) > (limit)) \ | |
283 { \ | |
284 const unsigned char *p0 = (p); \ | |
285 do { \ | |
286 p0--; \ | |
287 } while (p0 >= limit && ! CHAR_HEAD_P (*p0)); \ | |
288 (p) = (BYTES_BY_CHAR_HEAD (*p0) == (p) - p0) ? p0 : (p) - 1; \ | |
289 } \ | |
290 } while (0) | |
88363 | 291 |
292 /* Return the character code of character whose multibyte form is at | |
106185
f2cea199b0c4
* character.h (STRING_CHAR, STRING_CHAR_AND_LENGTH): Remove
Andreas Schwab <schwab@linux-m68k.org>
parents:
105872
diff
changeset
|
293 P. */ |
88363 | 294 |
106185
f2cea199b0c4
* character.h (STRING_CHAR, STRING_CHAR_AND_LENGTH): Remove
Andreas Schwab <schwab@linux-m68k.org>
parents:
105872
diff
changeset
|
295 #define STRING_CHAR(p) \ |
88363 | 296 (!((p)[0] & 0x80) \ |
297 ? (p)[0] \ | |
298 : ! ((p)[0] & 0x20) \ | |
299 ? (((((p)[0] & 0x1F) << 6) \ | |
300 | ((p)[1] & 0x3F)) \ | |
301 + (((unsigned char) (p)[0]) < 0xC2 ? 0x3FFF80 : 0)) \ | |
302 : ! ((p)[0] & 0x10) \ | |
303 ? ((((p)[0] & 0x0F) << 12) \ | |
304 | (((p)[1] & 0x3F) << 6) \ | |
305 | ((p)[2] & 0x3F)) \ | |
89180
1d29c2b108e6
(CHAR_STRING): Call char_string if C is greater than
Kenichi Handa <handa@m17n.org>
parents:
89053
diff
changeset
|
306 : string_char ((p), NULL, NULL)) |
88363 | 307 |
308 | |
98025
d4e07000ca4c
Fix spelling and wording of comments.
Eli Zaretskii <eliz@gnu.org>
parents:
97817
diff
changeset
|
309 /* Like STRING_CHAR, but set ACTUAL_LEN to the length of multibyte |
106185
f2cea199b0c4
* character.h (STRING_CHAR, STRING_CHAR_AND_LENGTH): Remove
Andreas Schwab <schwab@linux-m68k.org>
parents:
105872
diff
changeset
|
310 form. */ |
88363 | 311 |
106185
f2cea199b0c4
* character.h (STRING_CHAR, STRING_CHAR_AND_LENGTH): Remove
Andreas Schwab <schwab@linux-m68k.org>
parents:
105872
diff
changeset
|
312 #define STRING_CHAR_AND_LENGTH(p, actual_len) \ |
88363 | 313 (!((p)[0] & 0x80) \ |
314 ? ((actual_len) = 1, (p)[0]) \ | |
315 : ! ((p)[0] & 0x20) \ | |
316 ? ((actual_len) = 2, \ | |
317 (((((p)[0] & 0x1F) << 6) \ | |
318 | ((p)[1] & 0x3F)) \ | |
319 + (((unsigned char) (p)[0]) < 0xC2 ? 0x3FFF80 : 0))) \ | |
320 : ! ((p)[0] & 0x10) \ | |
321 ? ((actual_len) = 3, \ | |
322 ((((p)[0] & 0x0F) << 12) \ | |
323 | (((p)[1] & 0x3F) << 6) \ | |
324 | ((p)[2] & 0x3F))) \ | |
89180
1d29c2b108e6
(CHAR_STRING): Call char_string if C is greater than
Kenichi Handa <handa@m17n.org>
parents:
89053
diff
changeset
|
325 : string_char ((p), NULL, &actual_len)) |
88363 | 326 |
327 | |
98025
d4e07000ca4c
Fix spelling and wording of comments.
Eli Zaretskii <eliz@gnu.org>
parents:
97817
diff
changeset
|
328 /* Like STRING_CHAR, but advance P to the end of multibyte form. */ |
88363 | 329 |
330 #define STRING_CHAR_ADVANCE(p) \ | |
331 (!((p)[0] & 0x80) \ | |
332 ? *(p)++ \ | |
333 : ! ((p)[0] & 0x20) \ | |
334 ? ((p) += 2, \ | |
335 ((((p)[-2] & 0x1F) << 6) \ | |
336 | ((p)[-1] & 0x3F) \ | |
89483 | 337 | ((unsigned char) ((p)[-2]) < 0xC2 ? 0x3FFF80 : 0))) \ |
88363 | 338 : ! ((p)[0] & 0x10) \ |
339 ? ((p) += 3, \ | |
340 ((((p)[-3] & 0x0F) << 12) \ | |
341 | (((p)[-2] & 0x3F) << 6) \ | |
342 | ((p)[-1] & 0x3F))) \ | |
89180
1d29c2b108e6
(CHAR_STRING): Call char_string if C is greater than
Kenichi Handa <handa@m17n.org>
parents:
89053
diff
changeset
|
343 : string_char ((p), &(p), NULL)) |
88363 | 344 |
345 | |
346 /* Fetch the "next" character from Lisp string STRING at byte position | |
347 BYTEIDX, character position CHARIDX. Store it into OUTPUT. | |
348 | |
349 All the args must be side-effect-free. | |
350 BYTEIDX and CHARIDX must be lvalues; | |
351 we increment them past the character fetched. */ | |
352 | |
353 #define FETCH_STRING_CHAR_ADVANCE(OUTPUT, STRING, CHARIDX, BYTEIDX) \ | |
91620
acb32817f0e8
Use "do...while (0)", not "if (1)..else" in macro definitions.
Jan Djärv <jan.h.d@swipnet.se>
parents:
91444
diff
changeset
|
354 do \ |
88363 | 355 { \ |
356 CHARIDX++; \ | |
357 if (STRING_MULTIBYTE (STRING)) \ | |
358 { \ | |
91805
c330bf7419fd
(FETCH_STRING_CHAR_ADVANCE, FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE)
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
91620
diff
changeset
|
359 unsigned char *ptr = &SDATA (STRING)[BYTEIDX]; \ |
88363 | 360 int len; \ |
361 \ | |
106185
f2cea199b0c4
* character.h (STRING_CHAR, STRING_CHAR_AND_LENGTH): Remove
Andreas Schwab <schwab@linux-m68k.org>
parents:
105872
diff
changeset
|
362 OUTPUT = STRING_CHAR_AND_LENGTH (ptr, len); \ |
88363 | 363 BYTEIDX += len; \ |
364 } \ | |
365 else \ | |
91805
c330bf7419fd
(FETCH_STRING_CHAR_ADVANCE, FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE)
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
91620
diff
changeset
|
366 { \ |
c330bf7419fd
(FETCH_STRING_CHAR_ADVANCE, FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE)
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
91620
diff
changeset
|
367 OUTPUT = SREF (STRING, BYTEIDX); \ |
c330bf7419fd
(FETCH_STRING_CHAR_ADVANCE, FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE)
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
91620
diff
changeset
|
368 BYTEIDX++; \ |
c330bf7419fd
(FETCH_STRING_CHAR_ADVANCE, FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE)
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
91620
diff
changeset
|
369 } \ |
88363 | 370 } \ |
91620
acb32817f0e8
Use "do...while (0)", not "if (1)..else" in macro definitions.
Jan Djärv <jan.h.d@swipnet.se>
parents:
91444
diff
changeset
|
371 while (0) |
88363 | 372 |
98025
d4e07000ca4c
Fix spelling and wording of comments.
Eli Zaretskii <eliz@gnu.org>
parents:
97817
diff
changeset
|
373 /* Like FETCH_STRING_CHAR_ADVANCE, but return a multibyte character |
d4e07000ca4c
Fix spelling and wording of comments.
Eli Zaretskii <eliz@gnu.org>
parents:
97817
diff
changeset
|
374 even if STRING is unibyte. */ |
89053
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
375 |
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
376 #define FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE(OUTPUT, STRING, CHARIDX, BYTEIDX) \ |
91620
acb32817f0e8
Use "do...while (0)", not "if (1)..else" in macro definitions.
Jan Djärv <jan.h.d@swipnet.se>
parents:
91444
diff
changeset
|
377 do \ |
89053
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
378 { \ |
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
379 CHARIDX++; \ |
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
380 if (STRING_MULTIBYTE (STRING)) \ |
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
381 { \ |
91805
c330bf7419fd
(FETCH_STRING_CHAR_ADVANCE, FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE)
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
91620
diff
changeset
|
382 unsigned char *ptr = &SDATA (STRING)[BYTEIDX]; \ |
89053
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
383 int len; \ |
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
384 \ |
106185
f2cea199b0c4
* character.h (STRING_CHAR, STRING_CHAR_AND_LENGTH): Remove
Andreas Schwab <schwab@linux-m68k.org>
parents:
105872
diff
changeset
|
385 OUTPUT = STRING_CHAR_AND_LENGTH (ptr, len); \ |
89053
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
386 BYTEIDX += len; \ |
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
387 } \ |
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
388 else \ |
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
389 { \ |
91805
c330bf7419fd
(FETCH_STRING_CHAR_ADVANCE, FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE)
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
91620
diff
changeset
|
390 OUTPUT = SREF (STRING, BYTEIDX); \ |
c330bf7419fd
(FETCH_STRING_CHAR_ADVANCE, FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE)
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
91620
diff
changeset
|
391 BYTEIDX++; \ |
89053
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
392 MAKE_CHAR_MULTIBYTE (OUTPUT); \ |
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
393 } \ |
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
394 } \ |
91620
acb32817f0e8
Use "do...while (0)", not "if (1)..else" in macro definitions.
Jan Djärv <jan.h.d@swipnet.se>
parents:
91444
diff
changeset
|
395 while (0) |
89053
06a2cbbeaee9
(LEADING_CODE_LATIN_1_MIN)
Kenichi Handa <handa@m17n.org>
parents:
89038
diff
changeset
|
396 |
88363 | 397 |
98025
d4e07000ca4c
Fix spelling and wording of comments.
Eli Zaretskii <eliz@gnu.org>
parents:
97817
diff
changeset
|
398 /* Like FETCH_STRING_CHAR_ADVANCE, but assumes STRING is multibyte. */ |
88363 | 399 |
400 #define FETCH_STRING_CHAR_ADVANCE_NO_CHECK(OUTPUT, STRING, CHARIDX, BYTEIDX) \ | |
91620
acb32817f0e8
Use "do...while (0)", not "if (1)..else" in macro definitions.
Jan Djärv <jan.h.d@swipnet.se>
parents:
91444
diff
changeset
|
401 do \ |
88363 | 402 { \ |
91805
c330bf7419fd
(FETCH_STRING_CHAR_ADVANCE, FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE)
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
91620
diff
changeset
|
403 unsigned char *ptr = &SDATA (STRING)[BYTEIDX]; \ |
88363 | 404 int len; \ |
405 \ | |
106185
f2cea199b0c4
* character.h (STRING_CHAR, STRING_CHAR_AND_LENGTH): Remove
Andreas Schwab <schwab@linux-m68k.org>
parents:
105872
diff
changeset
|
406 OUTPUT = STRING_CHAR_AND_LENGTH (ptr, len); \ |
88363 | 407 BYTEIDX += len; \ |
408 CHARIDX++; \ | |
409 } \ | |
91620
acb32817f0e8
Use "do...while (0)", not "if (1)..else" in macro definitions.
Jan Djärv <jan.h.d@swipnet.se>
parents:
91444
diff
changeset
|
410 while (0) |
88363 | 411 |
412 | |
98025
d4e07000ca4c
Fix spelling and wording of comments.
Eli Zaretskii <eliz@gnu.org>
parents:
97817
diff
changeset
|
413 /* Like FETCH_STRING_CHAR_ADVANCE, but fetch character from the current |
88363 | 414 buffer. */ |
415 | |
416 #define FETCH_CHAR_ADVANCE(OUTPUT, CHARIDX, BYTEIDX) \ | |
91620
acb32817f0e8
Use "do...while (0)", not "if (1)..else" in macro definitions.
Jan Djärv <jan.h.d@swipnet.se>
parents:
91444
diff
changeset
|
417 do \ |
88363 | 418 { \ |
419 CHARIDX++; \ | |
420 if (!NILP (current_buffer->enable_multibyte_characters)) \ | |
421 { \ | |
422 unsigned char *ptr = BYTE_POS_ADDR (BYTEIDX); \ | |
423 int len; \ | |
424 \ | |
106185
f2cea199b0c4
* character.h (STRING_CHAR, STRING_CHAR_AND_LENGTH): Remove
Andreas Schwab <schwab@linux-m68k.org>
parents:
105872
diff
changeset
|
425 OUTPUT= STRING_CHAR_AND_LENGTH (ptr, len); \ |
88363 | 426 BYTEIDX += len; \ |
427 } \ | |
428 else \ | |
429 { \ | |
430 OUTPUT = *(BYTE_POS_ADDR (BYTEIDX)); \ | |
431 BYTEIDX++; \ | |
432 } \ | |
433 } \ | |
91620
acb32817f0e8
Use "do...while (0)", not "if (1)..else" in macro definitions.
Jan Djärv <jan.h.d@swipnet.se>
parents:
91444
diff
changeset
|
434 while (0) |
88363 | 435 |
436 | |
98025
d4e07000ca4c
Fix spelling and wording of comments.
Eli Zaretskii <eliz@gnu.org>
parents:
97817
diff
changeset
|
437 /* Like FETCH_CHAR_ADVANCE, but assumes the current buffer is multibyte. */ |
88363 | 438 |
439 #define FETCH_CHAR_ADVANCE_NO_CHECK(OUTPUT, CHARIDX, BYTEIDX) \ | |
91620
acb32817f0e8
Use "do...while (0)", not "if (1)..else" in macro definitions.
Jan Djärv <jan.h.d@swipnet.se>
parents:
91444
diff
changeset
|
440 do \ |
88363 | 441 { \ |
442 unsigned char *ptr = BYTE_POS_ADDR (BYTEIDX); \ | |
443 int len; \ | |
444 \ | |
106185
f2cea199b0c4
* character.h (STRING_CHAR, STRING_CHAR_AND_LENGTH): Remove
Andreas Schwab <schwab@linux-m68k.org>
parents:
105872
diff
changeset
|
445 OUTPUT = STRING_CHAR_AND_LENGTH (ptr, len); \ |
88363 | 446 BYTEIDX += len; \ |
447 CHARIDX++; \ | |
448 } \ | |
91620
acb32817f0e8
Use "do...while (0)", not "if (1)..else" in macro definitions.
Jan Djärv <jan.h.d@swipnet.se>
parents:
91444
diff
changeset
|
449 while (0) |
88363 | 450 |
451 | |
98025
d4e07000ca4c
Fix spelling and wording of comments.
Eli Zaretskii <eliz@gnu.org>
parents:
97817
diff
changeset
|
452 /* Increment the buffer byte position POS_BYTE of the current buffer to |
88363 | 453 the next character boundary. No range checking of POS. */ |
454 | |
455 #define INC_POS(pos_byte) \ | |
456 do { \ | |
457 unsigned char *p = BYTE_POS_ADDR (pos_byte); \ | |
458 pos_byte += BYTES_BY_CHAR_HEAD (*p); \ | |
459 } while (0) | |
460 | |
461 | |
98025
d4e07000ca4c
Fix spelling and wording of comments.
Eli Zaretskii <eliz@gnu.org>
parents:
97817
diff
changeset
|
462 /* Decrement the buffer byte position POS_BYTE of the current buffer to |
88363 | 463 the previous character boundary. No range checking of POS. */ |
464 | |
465 #define DEC_POS(pos_byte) \ | |
466 do { \ | |
467 unsigned char *p; \ | |
468 \ | |
469 pos_byte--; \ | |
470 if (pos_byte < GPT_BYTE) \ | |
91805
c330bf7419fd
(FETCH_STRING_CHAR_ADVANCE, FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE)
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
91620
diff
changeset
|
471 p = BEG_ADDR + pos_byte - BEG_BYTE; \ |
88363 | 472 else \ |
91805
c330bf7419fd
(FETCH_STRING_CHAR_ADVANCE, FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE)
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
91620
diff
changeset
|
473 p = BEG_ADDR + GAP_SIZE + pos_byte - BEG_BYTE;\ |
88363 | 474 while (!CHAR_HEAD_P (*p)) \ |
475 { \ | |
476 p--; \ | |
477 pos_byte--; \ | |
478 } \ | |
479 } while (0) | |
480 | |
481 /* Increment both CHARPOS and BYTEPOS, each in the appropriate way. */ | |
482 | |
483 #define INC_BOTH(charpos, bytepos) \ | |
484 do \ | |
485 { \ | |
486 (charpos)++; \ | |
487 if (NILP (current_buffer->enable_multibyte_characters)) \ | |
488 (bytepos)++; \ | |
489 else \ | |
490 INC_POS ((bytepos)); \ | |
491 } \ | |
492 while (0) | |
493 | |
494 | |
495 /* Decrement both CHARPOS and BYTEPOS, each in the appropriate way. */ | |
496 | |
497 #define DEC_BOTH(charpos, bytepos) \ | |
498 do \ | |
499 { \ | |
500 (charpos)--; \ | |
501 if (NILP (current_buffer->enable_multibyte_characters)) \ | |
502 (bytepos)--; \ | |
503 else \ | |
504 DEC_POS ((bytepos)); \ | |
505 } \ | |
506 while (0) | |
507 | |
508 | |
98025
d4e07000ca4c
Fix spelling and wording of comments.
Eli Zaretskii <eliz@gnu.org>
parents:
97817
diff
changeset
|
509 /* Increment the buffer byte position POS_BYTE of the current buffer to |
88363 | 510 the next character boundary. This macro relies on the fact that |
511 *GPT_ADDR and *Z_ADDR are always accessible and the values are | |
512 '\0'. No range checking of POS_BYTE. */ | |
513 | |
514 #define BUF_INC_POS(buf, pos_byte) \ | |
515 do { \ | |
516 unsigned char *p = BUF_BYTE_ADDRESS (buf, pos_byte); \ | |
517 pos_byte += BYTES_BY_CHAR_HEAD (*p); \ | |
518 } while (0) | |
519 | |
520 | |
98025
d4e07000ca4c
Fix spelling and wording of comments.
Eli Zaretskii <eliz@gnu.org>
parents:
97817
diff
changeset
|
521 /* Decrement the buffer byte position POS_BYTE of the current buffer to |
88363 | 522 the previous character boundary. No range checking of POS_BYTE. */ |
523 | |
524 #define BUF_DEC_POS(buf, pos_byte) \ | |
525 do { \ | |
526 unsigned char *p; \ | |
527 pos_byte--; \ | |
528 if (pos_byte < BUF_GPT_BYTE (buf)) \ | |
91805
c330bf7419fd
(FETCH_STRING_CHAR_ADVANCE, FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE)
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
91620
diff
changeset
|
529 p = BUF_BEG_ADDR (buf) + pos_byte - BEG_BYTE; \ |
88363 | 530 else \ |
91805
c330bf7419fd
(FETCH_STRING_CHAR_ADVANCE, FETCH_STRING_CHAR_AS_MULTIBYTE_ADVANCE)
Stefan Monnier <monnier@iro.umontreal.ca>
parents:
91620
diff
changeset
|
531 p = BUF_BEG_ADDR (buf) + BUF_GAP_SIZE (buf) + pos_byte - BEG_BYTE;\ |
88363 | 532 while (!CHAR_HEAD_P (*p)) \ |
533 { \ | |
534 p--; \ | |
535 pos_byte--; \ | |
536 } \ | |
537 } while (0) | |
538 | |
539 | |
89887
028a1f06f612
(LEADING_CODE_LATIN_1_MIN) (LEADING_CODE_LATIN_1_MAX): Delete these
Kenichi Handa <handa@m17n.org>
parents:
89692
diff
changeset
|
540 /* If C is a character to be unified with a Unicode character, return |
028a1f06f612
(LEADING_CODE_LATIN_1_MIN) (LEADING_CODE_LATIN_1_MAX): Delete these
Kenichi Handa <handa@m17n.org>
parents:
89692
diff
changeset
|
541 the unified Unicode character. */ |
028a1f06f612
(LEADING_CODE_LATIN_1_MIN) (LEADING_CODE_LATIN_1_MAX): Delete these
Kenichi Handa <handa@m17n.org>
parents:
89692
diff
changeset
|
542 |
99976
33f97c9e6889
(MAYBE_UNIFY_CHAR): Call maybe_unify_char instead of
Kenichi Handa <handa@m17n.org>
parents:
98025
diff
changeset
|
543 #define MAYBE_UNIFY_CHAR(c) \ |
33f97c9e6889
(MAYBE_UNIFY_CHAR): Call maybe_unify_char instead of
Kenichi Handa <handa@m17n.org>
parents:
98025
diff
changeset
|
544 do { \ |
33f97c9e6889
(MAYBE_UNIFY_CHAR): Call maybe_unify_char instead of
Kenichi Handa <handa@m17n.org>
parents:
98025
diff
changeset
|
545 if (c > MAX_UNICODE_CHAR && c <= MAX_5_BYTE_CHAR) \ |
33f97c9e6889
(MAYBE_UNIFY_CHAR): Call maybe_unify_char instead of
Kenichi Handa <handa@m17n.org>
parents:
98025
diff
changeset
|
546 { \ |
33f97c9e6889
(MAYBE_UNIFY_CHAR): Call maybe_unify_char instead of
Kenichi Handa <handa@m17n.org>
parents:
98025
diff
changeset
|
547 Lisp_Object val; \ |
33f97c9e6889
(MAYBE_UNIFY_CHAR): Call maybe_unify_char instead of
Kenichi Handa <handa@m17n.org>
parents:
98025
diff
changeset
|
548 val = CHAR_TABLE_REF (Vchar_unify_table, c); \ |
33f97c9e6889
(MAYBE_UNIFY_CHAR): Call maybe_unify_char instead of
Kenichi Handa <handa@m17n.org>
parents:
98025
diff
changeset
|
549 if (INTEGERP (val)) \ |
33f97c9e6889
(MAYBE_UNIFY_CHAR): Call maybe_unify_char instead of
Kenichi Handa <handa@m17n.org>
parents:
98025
diff
changeset
|
550 c = XINT (val); \ |
33f97c9e6889
(MAYBE_UNIFY_CHAR): Call maybe_unify_char instead of
Kenichi Handa <handa@m17n.org>
parents:
98025
diff
changeset
|
551 else if (! NILP (val)) \ |
33f97c9e6889
(MAYBE_UNIFY_CHAR): Call maybe_unify_char instead of
Kenichi Handa <handa@m17n.org>
parents:
98025
diff
changeset
|
552 c = maybe_unify_char (c, val); \ |
33f97c9e6889
(MAYBE_UNIFY_CHAR): Call maybe_unify_char instead of
Kenichi Handa <handa@m17n.org>
parents:
98025
diff
changeset
|
553 } \ |
33f97c9e6889
(MAYBE_UNIFY_CHAR): Call maybe_unify_char instead of
Kenichi Handa <handa@m17n.org>
parents:
98025
diff
changeset
|
554 } while (0) |
88363 | 555 |
88742
55e36a0cf0ee
(MAYBE_UNIFY_CHAR): Adjusted for the change of
Kenichi Handa <handa@m17n.org>
parents:
88545
diff
changeset
|
556 |
88363 | 557 /* Return the width of ASCII character C. The width is measured by |
98025
d4e07000ca4c
Fix spelling and wording of comments.
Eli Zaretskii <eliz@gnu.org>
parents:
97817
diff
changeset
|
558 how many columns C will occupy on the screen when displayed in the |
88363 | 559 current buffer. */ |
560 | |
561 #define ASCII_CHAR_WIDTH(c) \ | |
562 (c < 0x20 \ | |
563 ? (c == '\t' \ | |
564 ? XFASTINT (current_buffer->tab_width) \ | |
565 : (c == '\n' ? 0 : (NILP (current_buffer->ctl_arrow) ? 4 : 2))) \ | |
566 : (c < 0x7f \ | |
567 ? 1 \ | |
568 : ((NILP (current_buffer->ctl_arrow) ? 4 : 2)))) | |
569 | |
570 /* Return the width of character C. The width is measured by how many | |
98025
d4e07000ca4c
Fix spelling and wording of comments.
Eli Zaretskii <eliz@gnu.org>
parents:
97817
diff
changeset
|
571 columns C will occupy on the screen when displayed in the current |
88363 | 572 buffer. */ |
573 | |
574 #define CHAR_WIDTH(c) \ | |
575 (ASCII_CHAR_P (c) \ | |
576 ? ASCII_CHAR_WIDTH (c) \ | |
577 : XINT (CHAR_TABLE_REF (Vchar_width_table, c))) | |
578 | |
100794
434b92b258ff
(CHAR_VARIATION_SELECTOR_P): New macro.
Kenichi Handa <handa@m17n.org>
parents:
99976
diff
changeset
|
579 /* If C is a variation selector, return the index numnber of the |
434b92b258ff
(CHAR_VARIATION_SELECTOR_P): New macro.
Kenichi Handa <handa@m17n.org>
parents:
99976
diff
changeset
|
580 variation selector (1..256). Otherwise, return 0. */ |
434b92b258ff
(CHAR_VARIATION_SELECTOR_P): New macro.
Kenichi Handa <handa@m17n.org>
parents:
99976
diff
changeset
|
581 |
434b92b258ff
(CHAR_VARIATION_SELECTOR_P): New macro.
Kenichi Handa <handa@m17n.org>
parents:
99976
diff
changeset
|
582 #define CHAR_VARIATION_SELECTOR_P(c) \ |
434b92b258ff
(CHAR_VARIATION_SELECTOR_P): New macro.
Kenichi Handa <handa@m17n.org>
parents:
99976
diff
changeset
|
583 ((c) < 0xFE00 ? 0 \ |
434b92b258ff
(CHAR_VARIATION_SELECTOR_P): New macro.
Kenichi Handa <handa@m17n.org>
parents:
99976
diff
changeset
|
584 : (c) <= 0xFE0F ? (c) - 0xFE00 + 1 \ |
434b92b258ff
(CHAR_VARIATION_SELECTOR_P): New macro.
Kenichi Handa <handa@m17n.org>
parents:
99976
diff
changeset
|
585 : (c) < 0xE0100 ? 0 \ |
434b92b258ff
(CHAR_VARIATION_SELECTOR_P): New macro.
Kenichi Handa <handa@m17n.org>
parents:
99976
diff
changeset
|
586 : (c) <= 0xE01EF ? (c) - 0xE0100 + 17 \ |
434b92b258ff
(CHAR_VARIATION_SELECTOR_P): New macro.
Kenichi Handa <handa@m17n.org>
parents:
99976
diff
changeset
|
587 : 0) |
434b92b258ff
(CHAR_VARIATION_SELECTOR_P): New macro.
Kenichi Handa <handa@m17n.org>
parents:
99976
diff
changeset
|
588 |
434b92b258ff
(CHAR_VARIATION_SELECTOR_P): New macro.
Kenichi Handa <handa@m17n.org>
parents:
99976
diff
changeset
|
589 /* If C is a high surrogate, return 1. If C is a low surrogate, |
108933 | 590 return 0. Otherwise, return 0. */ |
100794
434b92b258ff
(CHAR_VARIATION_SELECTOR_P): New macro.
Kenichi Handa <handa@m17n.org>
parents:
99976
diff
changeset
|
591 |
434b92b258ff
(CHAR_VARIATION_SELECTOR_P): New macro.
Kenichi Handa <handa@m17n.org>
parents:
99976
diff
changeset
|
592 #define CHAR_SURROGATE_PAIR_P(c) \ |
434b92b258ff
(CHAR_VARIATION_SELECTOR_P): New macro.
Kenichi Handa <handa@m17n.org>
parents:
99976
diff
changeset
|
593 ((c) < 0xD800 ? 0 \ |
434b92b258ff
(CHAR_VARIATION_SELECTOR_P): New macro.
Kenichi Handa <handa@m17n.org>
parents:
99976
diff
changeset
|
594 : (c) <= 0xDBFF ? 1 \ |
434b92b258ff
(CHAR_VARIATION_SELECTOR_P): New macro.
Kenichi Handa <handa@m17n.org>
parents:
99976
diff
changeset
|
595 : (c) <= 0xDFFF ? 2 \ |
434b92b258ff
(CHAR_VARIATION_SELECTOR_P): New macro.
Kenichi Handa <handa@m17n.org>
parents:
99976
diff
changeset
|
596 : 0) |
434b92b258ff
(CHAR_VARIATION_SELECTOR_P): New macro.
Kenichi Handa <handa@m17n.org>
parents:
99976
diff
changeset
|
597 |
434b92b258ff
(CHAR_VARIATION_SELECTOR_P): New macro.
Kenichi Handa <handa@m17n.org>
parents:
99976
diff
changeset
|
598 |
109099 | 599 extern int char_resolve_modifier_mask (int); |
600 extern int char_string (unsigned, unsigned char *); | |
601 extern int string_char (const unsigned char *, | |
602 const unsigned char **, int *); | |
88363 | 603 |
109099 | 604 extern int translate_char (Lisp_Object, int c); |
605 extern int char_printable_p (int c); | |
606 extern void parse_str_as_multibyte (const unsigned char *, int, int *, | |
607 int *); | |
109555
05e7e7c46ff0
Use const, move declarations to header files.
Dan Nicolaescu <dann@ics.uci.edu>
parents:
109165
diff
changeset
|
608 extern int parse_str_to_multibyte (const unsigned char *, int); |
109099 | 609 extern int str_as_multibyte (unsigned char *, int, int, int *); |
610 extern int str_to_multibyte (unsigned char *, int, int); | |
611 extern int str_as_unibyte (unsigned char *, int); | |
612 extern EMACS_INT str_to_unibyte (const unsigned char *, unsigned char *, | |
613 EMACS_INT, int); | |
109555
05e7e7c46ff0
Use const, move declarations to header files.
Dan Nicolaescu <dann@ics.uci.edu>
parents:
109165
diff
changeset
|
614 extern int strwidth (const unsigned char *, int); |
109099 | 615 extern int c_string_width (const unsigned char *, int, int, int *, int *); |
616 extern int lisp_string_width (Lisp_Object, int, int *, int *); | |
88363 | 617 |
618 extern Lisp_Object Vprintable_chars; | |
619 | |
620 extern Lisp_Object Qcharacterp, Qauto_fill_chars; | |
621 extern Lisp_Object Vtranslation_table_vector; | |
622 extern Lisp_Object Vchar_width_table; | |
623 extern Lisp_Object Vchar_direction_table; | |
624 extern Lisp_Object Vchar_unify_table; | |
97817
f15876e66681
(Vunicode_category_table): Extern it.
Kenichi Handa <handa@m17n.org>
parents:
96250
diff
changeset
|
625 extern Lisp_Object Vunicode_category_table; |
88363 | 626 |
109099 | 627 extern Lisp_Object string_escape_byte8 (Lisp_Object); |
88545 | 628 |
88363 | 629 /* Return a translation table of id number ID. */ |
630 #define GET_TRANSLATION_TABLE(id) \ | |
631 (XCDR(XVECTOR(Vtranslation_table_vector)->contents[(id)])) | |
632 | |
633 /* A char-table for characters which may invoke auto-filling. */ | |
634 extern Lisp_Object Vauto_fill_chars; | |
635 | |
88915
94184802d0cc
(Vchar_script_table): Extern it.
Kenichi Handa <handa@m17n.org>
parents:
88873
diff
changeset
|
636 extern Lisp_Object Vchar_script_table; |
90402
69ac9cbd4be5
(Vscript_representative_chars): Extern it.
Kenichi Handa <handa@m17n.org>
parents:
90051
diff
changeset
|
637 extern Lisp_Object Vscript_representative_chars; |
88873
7d441bc35e9b
(TRAILING_CODE_P): New macro.
Kenichi Handa <handa@m17n.org>
parents:
88832
diff
changeset
|
638 |
88363 | 639 #define DEFSYM(sym, name) \ |
105872
89183b6f6e9d
* process.c (socket_options): Make it const.
Dan Nicolaescu <dann@ics.uci.edu>
parents:
105661
diff
changeset
|
640 do { (sym) = intern_c_string ((name)); staticpro (&(sym)); } while (0) |
88363 | 641 |
642 #endif /* EMACS_CHARACTER_H */ | |
89911 | 643 |
644 /* arch-tag: 4ef86004-2eff-4073-8cea-cfcbcf7188ac | |
645 (do not change this comment) */ |